Google leans on AI and genomics to help restore Australia’s giant kelp forests

As COP31 begins this week, Google is highlighting a conservation effort that uses artificial intelligence and genomics to slow the decline of giant kelp forests in Australia. The company said researchers working with Australia’s national science agency, CSIRO, are combining lab-based genetic work with AI analysis to support future kelp replanting that prioritizes heat tolerance.

Google’s update points to giant kelp as more than an underwater habitat. The company said kelp forests protect coastlines and provide shelter and ecosystem support, but are disappearing worldwide as oceans warm. In that context, restoring kelp is both an ecological and a climate resilience challenge.

The foundation for the project, Google said, is a set of two high-quality reference genomes for Australian giant kelp. Reference genomes are baseline genetic maps that allow scientists to compare new samples consistently, which in turn can help researchers track variation linked to traits like growth, survival, and stress tolerance.

Google said the next stage involved adapting a tool from its open-source AI ecosystem. Researchers trained a version of DeepVariant, an AI-based method for calling DNA variants from sequencing data, to analyze giant kelp DNA. DeepVariant is designed to translate sequencing reads into genetic information, helping researchers identify differences across individuals and populations.

According to Google, researchers used DeepVariant to sequence 600 wild kelp specimens collected from 40 sites along the Tasmanian coast. The company said those kelp populations have dropped to about 5% of historic levels, underscoring the scale of the loss that restoration efforts must overcome.

Google described how genetic analysis and partner lab studies are expected to feed into cross-breeding research aimed at producing kelp that can better withstand warmer conditions. The company said the long-term goal is to replant giant kelp forests with diverse specimens chosen for resilience rather than relying on a narrow genetic pool.

The work is part of Google Australia’s Digital Futures Initiative, a program the company frames around applying technology to practical public-benefit challenges. For Alphabet, the project also reflects a broader strategy shift toward demonstrating AI capabilities through environmental and scientific use cases, rather than limiting demonstrations to product features.

Sector context matters because kelp restoration depends on more than planting. As ocean temperatures increase, future survival may hinge on selecting genetic variation that maintains ecosystem functions under stress. That makes conservation genetics, including genomics-informed breeding, a critical bridge between lab research and field outcomes.

Google’s posting did not provide detail on timelines, expected planting sites, or field trial results. It also did not disclose specific measures of success, such as survival rates over time or quantified improvements in heat tolerance. The company’s emphasis is on the research groundwork and the genetic toolkit being built, rather than on reported restoration outcomes so far.

Looking ahead, the next milestones to watch are whether the genomics-informed breeding and replanting approach demonstrates measurable ecological recovery, and how quickly results from AI-assisted sequencing translate into durable kelp forest resilience. For regulators and scientists, the key question will be how effectively genetic diversity and heat-tolerant traits can be maintained at scale in real coastal conditions.